Decision Stump Feature Selection Based Mean Shift Brown Boost Map Reduce Clustering For Predictive Analytics With Big Data

• Published in: NeuroQuantology Vol. 20; no. 9; p. 5698
• Main Authors: Anita, M, Shakila, S
• Format: Journal Article
• Language: English
• Published: Bornova Izmir NeuroQuantology 01.01.2022
• Subjects: Big Data; Clustering; Data analysis; Decision trees; Error reduction; Feature selection; Spatial data
• ISSN: 1303-5150
• DOI: 10.14704/nq.2022.20.9.NQ44665

Big data refers to the generation of a huge volume of data continuously. Hence, analytics on such as large volume of data is becoming more complex regarding more time consumption and memory usage. With the aim of enhancing prediction accuracy by lesser time consumption, Decision Stump Feature Selection based Mean Shift Brown Boost Map Reduce Data Clustering (DSFS-MSBBMPDC) Technique is introduced for analyzing the spatial data to predict the future results. DSFS-MSBBMPDC technique consists of various procedures such as feature selection and clustering process to predictfuture results. First, the Otsuka-Ochiai decision stump Feature Selection was performed for choosing significant features. By one internal node, decision tree is linked to terminal node. After feature selection, the mean Shift steepest descent Brown Boost Map Reduce Data clustering process was performed to group input data to perform spatial data analysis. Brown Boost cluster combines the weak learner to form strong cluster. The prediction accuracy was increased as well as prediction error was reduced using the steepest descent function. The simulation is achieved by geographical dataset with various parameters by amount of features and amount of data. DSFS-MSBBMPDC improves performance compared with state-of-the-art works.